# Changelog of Site Audit: SEO, Broken Links & AI-Crawler Check (`gapforge/site-audit`) Actor

- **URL**: https://apify.com/gapforge/site-audit/changelog.md
- **Full Actor documentation**: https://apify.com/gapforge/site-audit.md

## Changelog

All notable changes to this Actor are recorded here. The format follows Keep a Changelog, and versions follow Semantic Versioning (MAJOR.MINOR; a breaking output change means a new MAJOR).

### \[1.0.11] - 2026-11-18

Build 11 of version 1.0: the output fields, the prices and the audit itself are unchanged.

#### Changed

- README: HTML tags in the issue-code table (for example `<title>`, `<h1>`, `<html lang="en">`) are shown as code, so the Store page no longer hides them.
- The **Maximum pages per site** field note is written in plain words.

### \[1.0.10] - 2026-11-16

Build 10 of version 1.0: the output fields, the prices and the audit itself are unchanged.

#### Changed

- More automated tests for how a rejected TLS certificate is reported. An expired certificate, one that is not trusted and one that does not match the host name each give `SITE_UNREACHABLE` plus, with security checks on, `TLS_INVALID` in the free site row, with no page charged. A connection that drops before any certificate arrives is reported as unreachable only, never as an invalid certificate.

### \[1.0] - 2026-11-15

#### Added

- Same-site crawl over plain HTTP from the start address and, optionally, from the site's sitemaps. robots.txt is followed (Allow and Disallow precedence, wildcards, Crawl-delay) for the user agent `GapForgeSiteAudit`.
- One `site` summary row per site: robots.txt, sitemaps, `llms.txt`, robots.txt rules for 12 AI crawlers, the http to https redirect, six security headers, the TLS certificate, and totals by severity.
- One `page` row per audited page: status, redirects, title, meta description, canonical, indexing directives, headings, language, hreflang, JSON-LD types and parse errors, images without alt, link counts, broken links, sitemap membership and depth.
- A fixed list of 50 documented issue codes, each with a severity, a message and a fix.
- Optional check of links to other sites (HEAD, then GET), with a per-site limit.
- Monitor mode: only pages with new or resolved issues since the last run are delivered; state is kept in a named key-value store whose name comes from the state key (lower-case letters and digits, with single dashes between them).
- Pay-per-event: `page-audit` charged after each page row is stored. The site summary row is free; unreachable sites and pages that do not answer add no page charge (a run still pays the small start fee). The run stops cleanly at the user's maximum charge.
- Dataset views Overview, Pages and Issues, and an output schema that links to them.
- A prefilled input that works out of the box (`https://books.toscrape.com/`, a public practice site for crawlers).
